Increased Concurrent SMTP Connections and STARTTLS Timeout After Upgrading to 8.0.2.3.20181024 (Doc ID 2482821.1)

Last updated on JULY 07, 2023

Applies to:

Symptoms

The following behavioral changes were observed after patching Messaging Server / MTA to update from 8.0.2.3.20180614 to 8.0.2.3.20181024:

• Increased number of concurrent SMTP connections

Number of concurrent connections to port 25 on the MTA bumped.

• Alarms on server response time

Started to get alarms from msprobe because it was taking longer than normal to connect to port 25.

The flag 'ct=600' is the key. That shows the connection lasted 10 minutes.  So they were stuck in STARTTLS for 10 minutes.

Connections which were previously working in a relatively normal/short length of time are now stuck for 10 minutes waiting for STARTTLS to complete.  That drives up the number of concurrent (mostly unproductive) connections.
Which puts us over the MAX_PROCS x MAX_CONNS limit. So the dispatcher is unable to accept any new connections. So clients time out.

• Additional Symptom: server flapping

The load-balancer (LB) see this as the server being unresponsive, takes it out of the VIP, which causes the load to drop the LB then sees it as responsive again, starts sending new connections to it, lather, rinse, repeat…

Could there be any change between Messaging Server 8.0.2.3.20180614 and 8.0.2.3.20181024 which might explain this behavior change?

Changes

Issue started after patching Messaging Server / MTA from 8.0.2.3.20180614 to 8.0.2.3.20181024...

Cause

My Oracle Support provides customers with access to over a million knowledge articles and a vibrant support community of peers and Oracle experts.